Add option to work with a subset of BBS data and fix passing the path…

… argument around in build_plans

NAMESPACE

@@ -11,6 +11,7 @@ export(collect_analyses)
 export(combine_subspecies)
 export(filter_bbs_species)
 export(filter_ts)
+export(get_bbs_route_region_data)
 export(get_cowley_lizards)
 export(get_cowley_snakes)
 export(get_default_data_path)

R/bbs_cleaning_functions.R

@@ -6,12 +6,13 @@
 #' @param start_yr num first year of time-series
 #' @param end_yr num last year of time-series
 #' @param min_num_yrs num minimum number of years of data between start_yr & end_yr
+#' @param selected_set optional, a subset of the BBS communities to use (to speed up development). As c(1:X)
 #' @inheritParams get_mtquad_data
 #' @return NULL
 #' @export
 
 prepare_bbs_ts_data <- function(start_yr = 1965, end_yr = 2017, min_num_yrs = 10,
-                                path = get_default_data_path()){
+                                path = get_default_data_path(), selected_set = NULL){
     bbs_data_tables <- import_retriever_data("breed-bird-survey", path = path)
 
     bbs_data <- bbs_data_tables$breed_bird_survey_weather %>%
@@ -48,6 +49,10 @@ prepare_bbs_ts_data <- function(start_yr = 1965, end_yr = 2017, min_num_yrs = 10
     }
     bbs_routes_regions_list = apply(bbs_routes_regions, MARGIN = 1, FUN = make_list)
 
+    if(!is.null(selected_set)) {
+        bbs_routes_regions_list = bbs_routes_regions_list[selected_set]
+    }
+
     lapply(bbs_routes_regions_list, FUN = subset_bbs_route_region_data, bbs_data_table = bbs_data, species_table = bbs_data_tables$breed_bird_survey_species, path = path)
 
 }
@@ -105,8 +110,6 @@ subset_bbs_route_region_data <- function(route_region, bbs_data_table, species_t
 
     saveRDS(this_bbs_result, file = file.path(storage_path, paste0("route", route, "region", region, ".Rds")) )
 
-    return()
-
 }
 
 

R/build_plans.R

@@ -65,6 +65,7 @@ build_analyses_plan <- function(methods, datasets, ...)
 #' @param data_path where to get the downloaded retriever datasets
 #' @param include_downloaded_data whether to also include downloadable datasets
 #' @param include_bbs_data whether to include BBS data
+#' @param bbs_subset optional, a subset of the BBS communities to use (to speed up development). As c(1:X)
 #' 
 #' @return a drake plan (i.e. a tibble) specifying the targets and commands 
 #'   for gathering datasets
@@ -73,7 +74,8 @@ build_analyses_plan <- function(methods, datasets, ...)
 #' 
 build_datasets_plan <- function(data_path = get_default_data_path(), 
                                 include_downloaded_data = FALSE,
-                                include_bbs_data = FALSE)
+                                include_bbs_data = FALSE,
+                                bbs_subset = NULL)
 {
     datasets <- drake::drake_plan(
         maizuru_data = get_maizuru_data(),
@@ -97,7 +99,12 @@ build_datasets_plan <- function(data_path = get_default_data_path(),
     }
 
     if (include_bbs_data) {
-        bbs_datasets = build_bbs_datasets_plan()
+        bbs_datasets = build_bbs_datasets_plan(data_path = data_path)
+
+        if(!is.null(bbs_subset)) {
+            bbs_datasets = bbs_datasets[bbs_subset, ]
+        }
+
         datasets <- datasets %>%
             dplyr::bind_rows(bbs_datasets)
     }
@@ -107,27 +114,27 @@ build_datasets_plan <- function(data_path = get_default_data_path(),
 
 #' @title Generate a Drake Plan for BBS Datasets
 #' 
-#' @param data_path path
+#' @param path path
 #' @param from_raw whether to re-prep BBS data
 #' 
 #' @return a drake plan (i.e. a tibble) specifying the targets and commands 
 #'   for gathering BBS datasets
 #' 
 #' @export
 #' 
-build_bbs_datasets_plan <- function(path = get_default_data_path())
+build_bbs_datasets_plan <- function(data_path = get_default_data_path())
 {
-    if(!file.exists(file.path(path, "breed-bird-survey-prepped", "routes_and_regions_table.csv"))) {
+    if(!file.exists(file.path(data_path, "breed-bird-survey-prepped", "routes_and_regions_table.csv"))) {
         prepare_bbs_ts_data()
     }
 
-    routes_and_regions = read.csv(file.path(path, "breed-bird-survey-prepped", "routes_and_regions_table.csv"), stringsAsFactors = F)
+    routes_and_regions = read.csv(file.path(data_path, "breed-bird-survey-prepped", "routes_and_regions_table.csv"), stringsAsFactors = F)
 
     routes_and_regions = routes_and_regions %>%
     dplyr::mutate(bcr = as.character(bcr), route = as.character(route))
 
     bbs_datasets <- drake::drake_plan(
-        bbs_data_rtrg = target(get_bbs_route_region_data(route, region, path = get_default_data_path()),
+        bbs_data_rtrg = target(get_bbs_route_region_data(route, region, path = !!data_path),
                                transform = map(route = !!rlang::syms(routes_and_regions$route),
                                                region = !!rlang::syms(routes_and_regions$bcr)
                                )